When you click on a website and see an ad for shoes you’ve been eyeing, it might seem straightforward: The advertiser paid the publisher to show you their product. But there are powerful middlemen—advertising agencies—that wield more market power than most people realize, and ultimately decide which ad you see.
New research from Amin Sayedi, Professor of Marketing and Michael G. Foster Endowed Fellow at the Foster School of Business, reveals that these agencies have grown so powerful that they can significantly reduce advertising revenue even for top publishers (like Google) in the online advertising market.
His recent Marketing Science paper, “Bid Coordination and Information Disclosure in Online Advertising Markets,” co-authored with Jason Choi of the University of Maryland, points to a counterintuitive strategy that publishers can employ to reclaim some of that power: withholding information.”
Who runs the online advertising ecosystem?
To understand the power dynamics at play, it helps to know who the key players are. In online advertising, “publishers” include any platform that displays ads. Google, Meta, Amazon, and even retailers like Walmart and Instacart are all publishers when they sell advertising space on their websites and apps.
These publishers sell opportunities to show ads to specific audiences at specific moments. When you search for “running shoes” on Google or scroll through Instagram, publishers are auctioning off the chance to reach you.
Enter the advertising agencies. Traditionally, agencies have served a valuable role: They help brands determine their messaging, identify target audiences, design campaigns, and then execute those strategies by purchasing ad space. For publishers, agencies are crucial partners because they bring together thousands of advertisers who want to reach customers.
While hundreds of advertising agencies exist, the industry is dominated by just a handful of massive holding companies like Omnicom, WPP, and Publicis. Each of these giants owns dozens or even hundreds of agencies under its umbrella. The holding companies aggregate demand across all these agencies through specialized “trading desks” that pool together all their clients’ advertising budgets to negotiate better prices with publishers.
For a long time, this was how the online advertising market worked: Agencies got better deals for their clients, and publishers got reliable, aggregated demand. But agencies operating as middlemen end up with considerable power in these advertising transactions, and that power doesn’t always benefit publishers.
How ad agencies aggregate power in the online advertising market
That dynamic became clear to Sayedi while working with Amazon’s advertising platform. The retail giant had built Amazon DSP (Demand Side Platform), a technology platform that advertising agencies use to buy ads programmatically across the web.
Sayedi, who teaches in Foster’s MSBA program, thought agencies would obviously benefit from the platform’s capabilities, “but some of the behavior was not exactly what I anticipated, meaning that they were not fully using this power that they had,” he says.
As he dug deeper, he realized that publishers were behaving in ways that forced the middleman to give up some of their power.
Around the same time, an economics paper revealed that agencies’ market power was far greater than anticipated, potentially impacting Google’s revenue by more than 10%.
“That’s a lot of money,” Sayedi says. “If they can hurt Google that much, then this is something that publishers should actively think about.”
The mechanism behind this power is what Sayedi calls “bid coordination.” Here’s how it works: Imagine the same agency represents both Beats and Bose, both premium headphone makers. Usually, these brands would set up their automated bidding strategies to compete aggressively for people interested in high-end headphones—each trying to outbid the other for that valuable audience, which drives up prices.
But if one agency represents both brands, it can coordinate their bidding strategies to eliminate that competition. The agency can split the ad opportunities between them, and neither has to compete. Both get those ad slots at much lower prices than they would otherwise.
For advertisers, this creates enormous value because they’re getting better returns on their ad spending. For agencies, it justifies their fees by demonstrating clear ROI to clients. But for publishers like Google, Meta, or Amazon, it means significantly reduced revenue from what should be competitive auctions.
Publishers withhold information to limit agency coordination
Sayedi’s research reveals a surprising strategy for publishers: intentionally restrict the information they share with agencies and advertisers.
The mechanism is straightforward.
“If you don’t share much information, you’re kind of forcing the agencies to give up on bid coordination, or maybe to lower how much bid coordination they can do,” Sayedi says.
Without detailed information about audiences and contexts, agencies can’t effectively coordinate bids between competing brands.
But there’s a risk: If the best ads aren’t shown to customers, that hurts publishers too. The key is identifying which information truly enables better ad targeting versus primarily enabling agency coordination. For example, customer gender is relevant for shoe ads but not for phone chargers. Whether someone uses an iPhone or an Android matters for accessory makers. Publishers must be strategic about what they reveal.
“Today, Google shares a lot less information with agencies and advertisers than what it was sharing like five years ago,” Sayedi says.
Why new publishers like Walmart must play by different rules
But not every publisher can play the game the way established publishers like Google do. Emerging publishers like Walmart and Instacart have to play by the agencies’ rules—at least for now.
“Even though Walmart is huge as a retailer, they are not huge as a publisher,” Sayedi says.
New publishers face a chicken-and-egg problem: They need to demonstrate that ads on their platform drive sales, but they need advertiser spending to gather that proof. Withholding information at this stage would be counterproductive.
“You have to establish yourself first, and you have to become a big player in the market first before you can change those dynamics,” Sayedi says.
In other words, you have to fully understand the rules before you can break them.
For emerging publishers, the strategy is clear: Collaborate closely with agencies, make information readily available, automate wherever possible, and prove your platform creates value. Only once you’ve achieved significant scale does the calculus shift toward strategic information control.
How research shapes Foster’s online advertising curriculum
Sayedi’s insights on advertising dynamics don’t just live in academic journals—they shape how he teaches Foster’s online advertising course in the MSBA program.
“I have only seven weeks, so I have to be very careful in what topics I select,” he says. “If someone from this class wants to work in online advertising after this, I want to make sure that they know the most important things.”
For Sayedi, the most important things include how publishers should think about pricing and information disclosure, how advertisers should approach bidding and buying, and how to measure advertising impact. “These are some of the main themes of the course,” he says, and they are all informed by research that examines the real tensions and power dynamics in the digital advertising ecosystem.
As the online advertising market continues to evolve, with new publishers entering and established players adjusting their strategies, understanding these power dynamics becomes increasingly critical. Information about audiences isn’t just data—it’s currency, and the choice of whether to spend it or save it can mean billions in revenue.
Read the research paper here.
Learn more about the Foster MSBA here.
Amin Sayedi is Professor of Marketing and Michael G. Foster Endowed Fellow at the Foster School of Business. He teaches in Foster’s 12-month Master of Science in Business Analytics (MSBA) program.